@tangle-network/agent-eval 0.118.0 → 0.118.2

package/dist/cli.js

@@ -5,9 +5,9 @@ import {
   runRpcBatch,
   runRpcOnce,
   startServer
-} from "./chunk-LTVG32KX.js";
+} from "./chunk-JTMFT5QZ.js";
 import "./chunk-NJC7U437.js";
-import "./chunk-VCTY3W6J.js";
+import "./chunk-MLHRFWQK.js";
 import "./chunk-VI2UW6B6.js";
 import "./chunk-PC4UYEBM.js";
 import "./chunk-ONWEPEDO.js";

package/dist/contract/index.d.ts

@@ -617,6 +617,13 @@ declare class CostLedger {
     private ensureCostLimitPersisted;
     private appendEvent;
 }
+/** Public callback surface for a shared cost ledger.
+ *
+ * Declaration bundles may expose this type through multiple package subpaths.
+ * Keeping callback contracts structural lets those subpaths compose while the
+ * concrete {@link CostLedger} retains its private durable state.
+ */
+type CostLedgerHandle = Pick<CostLedger, keyof CostLedger>;
 
 interface Scenario$1 {
     id: string;
@@ -673,7 +680,7 @@ interface JudgeInput {
     turns: TurnResult[];
     artifacts: CollectedArtifacts;
     /** Shared ledger for paid built-in judges. Direct calls default to an uncapped ledger. */
-    costLedger?: CostLedger;
+    costLedger?: CostLedgerHandle;
     costPhase?: string;
     costTags?: Record<string, string>;
     signal?: AbortSignal;
@@ -1389,7 +1396,7 @@ interface JudgeConfig<TArtifact, TScenario extends Scenario = Scenario> {
         scenario: TScenario;
         signal: AbortSignal;
         /** Shared run spend account and receipt attribution phase. */
-        costLedger?: CostLedger;
+        costLedger?: CostLedgerHandle;
         costPhase?: string;
         costTags?: Record<string, string>;
     }): JudgeScore | Promise<JudgeScore>;
@@ -1573,7 +1580,7 @@ interface ProposeContext<TFindings = unknown> {
      *  reflection may ignore it. See {@link ParetoParent}. */
     paretoParents?: ParetoParent[];
     /** Shared run spend account and receipt attribution phase. */
-    costLedger?: CostLedger;
+    costLedger?: CostLedgerHandle;
     costPhase?: string;
     /** FIREWALL (non-negotiable): the held-out judge is write-only — its verdicts
      *  score the chosen output and gate promotion, and are NEVER an input to
@@ -1657,7 +1664,7 @@ interface GateContext<TArtifact, TScenario extends Scenario> {
         baseline: number;
     };
     /** Shared run spend account and receipt attribution phase. */
-    costLedger?: CostLedger;
+    costLedger?: CostLedgerHandle;
     costPhase?: string;
     signal: AbortSignal;
 }
@@ -1999,7 +2006,7 @@ interface RunCampaignOptions<TScenario extends Scenario, TArtifact> {
     costCeiling?: number;
     /** Shared spend account. Improvement loops pass one ledger through every
      *  campaign so the ceiling and returned total are run-wide. */
-    costLedger?: CostLedger;
+    costLedger?: CostLedgerHandle;
     /** Attribution label for receipts recorded by this campaign. */
     costPhase?: string;
     /** Max concurrent cells. Default 2. */
@@ -2196,7 +2203,7 @@ interface RunOptimizationBaseOptions<TScenario extends Scenario, TArtifact> exte
         }>;
         history: GenerationRecord[];
         /** Shared run spend account and receipt attribution phase. */
-        costLedger?: CostLedger;
+        costLedger?: CostLedgerHandle;
         costPhase?: string;
     }) => Promise<unknown[]>;
 }
@@ -2325,7 +2332,7 @@ interface ReferenceEquivalenceJudgeOptions {
     /** Used only by the direct-call adapter. */
     signal?: AbortSignal;
     /** Optional receipt destination for direct calls; campaigns supply their own. */
-    costLedger?: CostLedger;
+    costLedger?: CostLedgerHandle;
 }
 interface ReferenceEquivalenceJudgeResult extends LlmCallMetadata {
     kind: 'reference-equivalence';
@@ -2414,7 +2421,7 @@ interface GepaProposerOptions {
     /** Model that performs the reflection. */
     model: string;
     /** Optional ledger for direct proposer use. Campaign context takes precedence. */
-    costLedger?: CostLedger;
+    costLedger?: CostLedgerHandle;
     /** What is being optimized — appears in the reflection prompt for orientation. */
     target: string;
     /** Surface-specific mutation levers offered to the model. */
@@ -3860,7 +3867,7 @@ interface SelfImproveResult<TScenario extends Scenario, TArtifact> {
 declare class SelfImproveRunError extends Error {
     readonly cost: CostLedgerSummary;
     readonly receipts: CostReceipt[];
-    constructor(cause: unknown, ledger: CostLedger);
+    constructor(cause: unknown, ledger: CostLedgerHandle);
 }
 /**
  * One-shot self-improvement loop. See module docstring for defaults +
@@ -4829,4 +4836,4 @@ interface FromOtelSpansOptions {
 }
 declare function fromOtelSpans(opts: FromOtelSpansOptions): RunRecord[];
 
-export { type AgentEvalAgent, type AgentEvalEvaluateOptions, type AgentEvalImproveOptions, type AgentTraceContributor, type AgentTraceContributorType, type AgentTraceConversation, type AgentTraceFile, type AgentTraceIndex, type AgentTraceRange, type AgentTraceRecord, type AnalystFinding, type AnalyzeRunsOptions, type AuthoringProvenance, type AxisEvidence, type AxisVerdict, type BuildEvidenceVectorOptions, type CampaignAggregates, type CampaignArtifactWriter, type CampaignCellResult, type CampaignCostMeter, type CampaignResult, type CampaignStorage, type CampaignTraceWriter, type ChatClient, type CodeAgentSessionDiagnostic, type CodeAgentSessionIntakeOptions, type CodeAgentSessionIntakeResult, type CodeAgentSessionMetrics, type CodeAgentSessionSource, type CodeSurface, type CostProvenanceSummary, type CreateChatClientOpts, type DefaultAnalystRegistryOptions, type DefaultProductionGateOptions, type DefineAgentEvalOptions, type DefinedAgentEval, type DeploymentOutcome, type DispatchFn as Dispatch, type DispatchContext, type EvalCellScoreDelta, type EvalDimensionDelta, type EvalGenerationDiff, type EvalReportingSuiteInput, type EvalReportingSuiteOptions, type EvalReportingSuiteResult, type EvalRunDiff, type EvidenceVector, type EvolutionaryProposerOptions, type ExecutionInsight, type ExecutionReport, type FailureClusterInsight, type FeedbackTableMeta, type FeedbackTableRow, FileSystemOutcomeStore, type FileSystemOutcomeStoreOptions, type FromFeedbackTableOptions, type FromFeedbackTableResult, type FromOtelSpansOptions, type FromRunRecordDirOptions, type FromRunRecordDirResult, type Gate, type GateContext, type GateDecision, type GateResult, type GenerationCandidate, type GenerationRecord, type GepaProposerOptions, type HeldOutGateOptions, type HostedTenant, InMemoryOutcomeStore, type InsightReport, type InterRaterInsight, type JudgeConfig, type JudgeDimension, type JudgeInsight, type JudgeScore, type LiftInsight, type MutableSurface, type Mutator, type ObjectiveSource, type OptimizationProposer, type OptimizerConfig, type OutcomeCorrelationInsight, type OutcomeStore, type ParetoSignificanceGateOptions, type ParsedCodeAgentJsonl, type PartitionByAuthoringModelResult, type PromotionObjective, type PromotionPolicy, REFERENCE_EQUIVALENCE_INPUT_LIMITS, REFERENCE_EQUIVALENCE_JUDGE_VERSION, type Recommendation, type ReferenceEquivalenceJudgeInput, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceJudgeResult, type ReferenceEquivalenceScenario, type ReleaseSummary, type RunCampaignOptions, type RunEvalOptions, type RunImprovementLoopOptions, type RunImprovementLoopResult, type RunRecordRejection, type ScalarDistribution, type Scenario, type SelfImproveBudget, type SelfImproveLlm, type SelfImproveOptions, type SelfImproveProgressEvent, type SelfImproveResult, SelfImproveRunError, type SessionScript, type SummarizeExecutionOptions, type SurfaceProposer, type TokenUsageInsight, analyzeRuns, buildDefaultAnalystRegistry, buildEvidenceVector, composeGate, createChatClient, createReferenceEquivalenceJudge, defaultProductionGate, defineAgentEval, diffGenerations, diffRunBaselineToWinner, diffRuns, evalReportingSuite, evolutionaryProposer, fromClaudeCodeSession, fromCodexSession, fromFeedbackTable, fromKimiCodeSession, fromOpenCodeSession, fromOtelSpans, fromPiSession, fromPigraphSession, fromRunRecordDir, fsCampaignStorage, gepaProposer, heldOutGate, inMemoryCampaignStorage, paretoPolicy, paretoSignificanceGate, parseAgentTrace, parseCodeAgentJsonl, partitionRunsByAuthoringModel, runCampaign, runEval, runImprovementLoop, runReferenceEquivalenceJudge, selfImprove, summarizeExecution };
+export { type AgentEvalAgent, type AgentEvalEvaluateOptions, type AgentEvalImproveOptions, type AgentTraceContributor, type AgentTraceContributorType, type AgentTraceConversation, type AgentTraceFile, type AgentTraceIndex, type AgentTraceRange, type AgentTraceRecord, type AnalystFinding, type AnalyzeRunsOptions, type AuthoringProvenance, type AxisEvidence, type AxisVerdict, type BuildEvidenceVectorOptions, type CampaignAggregates, type CampaignArtifactWriter, type CampaignCellResult, type CampaignCostMeter, type CampaignResult, type CampaignStorage, type CampaignTraceWriter, type ChatClient, type CodeAgentSessionDiagnostic, type CodeAgentSessionIntakeOptions, type CodeAgentSessionIntakeResult, type CodeAgentSessionMetrics, type CodeAgentSessionSource, type CodeSurface, type CostLedgerHandle, type CostProvenanceSummary, type CreateChatClientOpts, type DefaultAnalystRegistryOptions, type DefaultProductionGateOptions, type DefineAgentEvalOptions, type DefinedAgentEval, type DeploymentOutcome, type DispatchFn as Dispatch, type DispatchContext, type EvalCellScoreDelta, type EvalDimensionDelta, type EvalGenerationDiff, type EvalReportingSuiteInput, type EvalReportingSuiteOptions, type EvalReportingSuiteResult, type EvalRunDiff, type EvidenceVector, type EvolutionaryProposerOptions, type ExecutionInsight, type ExecutionReport, type FailureClusterInsight, type FeedbackTableMeta, type FeedbackTableRow, FileSystemOutcomeStore, type FileSystemOutcomeStoreOptions, type FromFeedbackTableOptions, type FromFeedbackTableResult, type FromOtelSpansOptions, type FromRunRecordDirOptions, type FromRunRecordDirResult, type Gate, type GateContext, type GateDecision, type GateResult, type GenerationCandidate, type GenerationRecord, type GepaProposerOptions, type HeldOutGateOptions, type HostedTenant, InMemoryOutcomeStore, type InsightReport, type InterRaterInsight, type JudgeConfig, type JudgeDimension, type JudgeInsight, type JudgeScore, type LiftInsight, type MutableSurface, type Mutator, type ObjectiveSource, type OptimizationProposer, type OptimizerConfig, type OutcomeCorrelationInsight, type OutcomeStore, type ParetoSignificanceGateOptions, type ParsedCodeAgentJsonl, type PartitionByAuthoringModelResult, type PromotionObjective, type PromotionPolicy, REFERENCE_EQUIVALENCE_INPUT_LIMITS, REFERENCE_EQUIVALENCE_JUDGE_VERSION, type Recommendation, type ReferenceEquivalenceJudgeInput, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceJudgeResult, type ReferenceEquivalenceScenario, type ReleaseSummary, type RunCampaignOptions, type RunEvalOptions, type RunImprovementLoopOptions, type RunImprovementLoopResult, type RunRecordRejection, type ScalarDistribution, type Scenario, type SelfImproveBudget, type SelfImproveLlm, type SelfImproveOptions, type SelfImproveProgressEvent, type SelfImproveResult, SelfImproveRunError, type SessionScript, type SummarizeExecutionOptions, type SurfaceProposer, type TokenUsageInsight, analyzeRuns, buildDefaultAnalystRegistry, buildEvidenceVector, composeGate, createChatClient, createReferenceEquivalenceJudge, defaultProductionGate, defineAgentEval, diffGenerations, diffRunBaselineToWinner, diffRuns, evalReportingSuite, evolutionaryProposer, fromClaudeCodeSession, fromCodexSession, fromFeedbackTable, fromKimiCodeSession, fromOpenCodeSession, fromOtelSpans, fromPiSession, fromPigraphSession, fromRunRecordDir, fsCampaignStorage, gepaProposer, heldOutGate, inMemoryCampaignStorage, paretoPolicy, paretoSignificanceGate, parseAgentTrace, parseCodeAgentJsonl, partitionRunsByAuthoringModel, runCampaign, runEval, runImprovementLoop, runReferenceEquivalenceJudge, selfImprove, summarizeExecution };

package/dist/contract/index.js

@@ -13,7 +13,7 @@ import {
 import {
   analyzeRuns,
   summarizeExecution
-} from "../chunk-SVH2ANFD.js";
+} from "../chunk-7V3QDWNL.js";
 import {
   REFERENCE_EQUIVALENCE_INPUT_LIMITS,
   REFERENCE_EQUIVALENCE_JUDGE_VERSION,
@@ -34,20 +34,20 @@ import {
   runReferenceEquivalenceJudge,
   surfaceContentHash,
   surfaceHash
-} from "../chunk-YLMUS4MM.js";
+} from "../chunk-I42AAGHL.js";
 import {
   createRunCostLedger,
   fsCampaignStorage,
   inMemoryCampaignStorage,
   resolveRunDir,
   runCampaign
-} from "../chunk-IDZTTFRR.js";
+} from "../chunk-EGNRE7VA.js";
 import {
   buildDefaultAnalystRegistry,
   createChatClient
-} from "../chunk-33JA4TFA.js";
+} from "../chunk-J23QK2U5.js";
 import "../chunk-HHWE3POT.js";
-import "../chunk-OIIMMLRB.js";
+import "../chunk-DZBCDDIE.js";
 import {
   FileSystemOutcomeStore,
   InMemoryOutcomeStore
@@ -56,18 +56,18 @@ import "../chunk-ARU2PZFM.js";
 import "../chunk-NJC7U437.js";
 import "../chunk-DPZAEKA6.js";
 import "../chunk-PJQFMIOX.js";
-import "../chunk-VCTY3W6J.js";
+import "../chunk-MLHRFWQK.js";
 import "../chunk-VI2UW6B6.js";
 import {
   recordAggregateMeasurements,
   summarizeExecutionMeasurements
-} from "../chunk-KSDQVPLR.js";
+} from "../chunk-TVXPJJ7H.js";
 import "../chunk-PXE2VKMX.js";
 import "../chunk-ZET2UAYW.js";
 import "../chunk-GGE4NNQT.js";
 import {
   LLM_MODEL_ATTR_KEYS
-} from "../chunk-QKEGNI5B.js";
+} from "../chunk-BUBU3YHT.js";
 import "../chunk-PC4UYEBM.js";
 import {
   parseRunRecordSafe
@@ -76,7 +76,7 @@ import "../chunk-MA6HLL3S.js";
 import "../chunk-XJYR7XFV.js";
 import "../chunk-VSMTAMNK.js";
 import "../chunk-ONWEPEDO.js";
-import "../chunk-WDHBCA3M.js";
+import "../chunk-NW4AQBXD.js";
 import "../chunk-PZ5AY32C.js";
 
 // src/contract/self-improve.ts